def __init__(self,
coords: Tuple[int, int],
locale: str = 'en',
name: str = '',
sign: str = '',
company: str = '',
text: str = '',
date: str = '',
year_start: Union[int, str] = '',
year_end: Union[str, int] = ''):
super().__init__(name=name, coords=coords, locale=locale)
self.sign = sign
self.company = company
self.text = text
self.date = date
self.year_start = TextAnnotation.get_int_value(year_start)
self.year_end = TextAnnotation.get_int_value(year_end)
def get_citation_annotations(cls, text: str) -> \
Generator[CitationAnnotation, None, None]:
"""
Get citations containing "BGBl"
:param text: str
:return: yields dict
"""
for ptn in [
cls.CITATION_PTN_RE, cls.SECOND_CITATION_PTN_RE,
cls.CITATION_RANGE_PTN_RE
]:
for match in ptn.finditer(text):
capture = match.capturesdict()
date = ''.join(capture.get('date', ''))
if date:
try:
date = str(list(get_dates(date, 'de'))[0]['value'])
except:
pass
ant = CitationAnnotation(
coords=match.span(),
text=capture['text'][0],
paragraph=''.join(capture.get('paragraph', '')),
subparagraph=''.join(capture.get('subparagraph', '')),
letter=''.join(capture.get('letter', '')),
date=date,
part=capture['part'][0],
locale='de')
ant.article = TextAnnotation.get_int_value(''.join(
capture.get('article', '')))
ant.number = TextAnnotation.get_int_value(''.join(
capture.get('number', '')))
ant.sentence = TextAnnotation.get_int_value(''.join(
capture.get('sentence', '')))
page_range = ', '.join(capture['page'])
page = TextAnnotation.get_int_value(page_range)
if page:
ant.page = page
else:
ant.page_range = page_range
volume_str = ''.join(capture.get('number', ''))
volume = TextAnnotation.get_int_value(volume_str)
if volume:
ant.volume = volume
else:
ant.volume_str = volume_str
year_str = ', '.join(capture.get('year', ''))
year = TextAnnotation.get_int_value(year_str)
if year:
ant.year = year
else:
ant.year_str = year_str
yield ant
def get_acts_annotations(text: str) -> Generator[ActAnnotation, None, None]:
for match in ACT_PARTS_RE.finditer(text):
captures = match.capturesdict()
act_name = ''.join(captures.get('act_name') or [])
year_str = ''.join(captures.get('year') or [])
year = TextAnnotation.safe_cast(year_str, int)
act = ActAnnotation(act_name=act_name,
coords=match.span(),
section=''.join(captures.get('section') or []),
year=year,
ambiguous=act_name == 'Act',
text=''.join(captures.get('text') or []),
locale='en')
yield act
def get_geoentity_annotations(
text: str,
geo_config_list: List[DictionaryEntry],
priority: bool = False,
priority_by_id: bool = False,
text_languages: List[str] = None,
min_alias_len: int = geoentities_config.MIN_ALIAS_LEN,
prepared_alias_ban_list: Union[None, Dict[str, Tuple[
List[str], List[str]]]] = _ALIAS_BLACK_LIST_PREPARED,
simplified_normalization: bool = False
) -> Generator[GeoAnnotation, None, None]:
"See get_geoentities"
conflict_resolving_func = None
if priority_by_id:
conflict_resolving_func = conflicts_take_first_by_id
if priority:
conflict_resolving_func = conflicts_top_by_priority
dic_entries = find_dict_entities(
text,
geo_config_list,
conflict_resolving_func=conflict_resolving_func,
text_languages=text_languages,
min_alias_len=min_alias_len,
prepared_alias_ban_list=prepared_alias_ban_list,
simplified_normalization=simplified_normalization)
for ent in dic_entries:
ant = GeoAnnotation(coords=ent.coords)
if ent.entity[0]:
toponim = ent.entity[0] # type: DictionaryEntry
year = TextAnnotation.get_int_value(toponim.id)
if year:
ant.year = year
ant.name = toponim.name
yield ant